The Agent Skills Directory

Indirect Prompt Injection Surface: The skill defines a workflow where the agent ingests external data from GitHub issues to generate branch names and Pull Request content. This presents a potential surface for indirect prompt injection if an issue contains instructions intended to influence the agent's behavior.
Ingestion points: Untrusted data enters the agent context via GitHub issue metadata (number, description, slug) as described in SKILL.md.
Boundary markers: There are no explicit boundary markers or instructions to treat external data as untrusted within the SKILL.md file.
Capability inventory: The skill utilizes subprocess calls for git and gh (GitHub CLI) and involves file system operations (worktrees, .squad/ state) as outlined in SKILL.md.
Sanitization: No sanitization or validation steps are defined for the issue-derived variables used in shell commands.
Command Execution Patterns: The skill utilizes standard development tools to automate a version control workflow.
Pattern: Usage of git worktree for isolated working directories and npm link/pip install for local dependency linking are standard development practices aligned with the skill's primary purpose.

git-workflow